Traditional thermal detectors, devices that can see into the long wavelength infrared (λ~8-14 μm) without any cooling, are broadband devices, detecting light across the entire long-wave spectrum. These are extraordinarily useful, having many commercial, industrial, and military applications, especially in imaging, night vision, and thermometry. However, this technology is extremely limited when it comes to other applications such as chemical sensing and hyperspectral imaging. Here only a small fraction of the infrared spectrum is relevant and all other signals are noise to be avoided. Our group works extensively in the area of spectrally selective uncooled detection.
Spectral selectivity in a thermal detector can be achieved in many ways. Perhaps the simplest is to place a filter of some sort in front of a normal broadband device. However, this is a very poor design for high performance (unless one has the luxury of cryogenic cooling) because it produces a very low fundamental signal-to-noise ratio. The signal is limited to the acceptance band of the filter, but the background noise comes from all wavelengths and directions. One can also fabricate a detector with materials that absorb in only a limited range. Alternatively, the device can be patterned to enhance or suppress different spectral regions. One can achieve limited or narrowband detection by combining a detector plate with an optical cavity.
|A bolometer with electrostatically-tunable optical resonance, in concept…||…and realized.|
The figures above show an early (2006) example of an electrostatically tunable infrared detector, where the detector plate is embedded inside the top mirror of an optical cavity with a bottom mirror of approximately 100% reflectivity. The plate absorption was designed and demonstrated to complement the reflectivity of the top mirror such that light on resonance is absorbed perfectly while light off-resonance is rejected by the cavity or interacts only weakly with the detector plate. By applying a voltage (0-42 V) one could tune the received wavelength band from 8.7 μm to 11+ μm, as shown in the figure below. This device was the first tunable uncooled infrared detector, but it was still limited by the same noise issues that plague broadband detectors and did not fully utilize its spectral selectivity to reduce the radiation background.
This idea of a coupled cavity absorption has received a lot of recent attention as a perfect absorber, but it is actually an old idea. This type of cavity is not unique to thermal detection and has been used in telecommunications detectors and thermal detectors, among other devices. Note that since the detector interacts with only a small part of the spectrum, its radiation noise is lower than that of a blackbody detector. Thus coupled absorption may make an even more profound contribution to technology even beyond its use in perfect absorbers. The figure above shows a detector plate placed inside an optical cavity. A simulated spectrum of such a device at different voltages is shown below left; note that the detector plate interacts with the radiation background only on resonance, making the background noise much lower even for an uncooled device! The figure below right illustrates how the theoretical background noise limit changes as a function of the FWHM of the resonance.
|Simulated spectral absorption of a detector place inside an optical cavity.||The theoretical blackbody background noise limit, which changes with the resonance width of the detector.|
Below is shown our first attempt to demonstrate these concepts, driving the detectivity of an uncooled detector towards the background limit, and a measurement of the lower limit for the detectivity of this device. While it is still far from the background limit, this device has achieved the current record for the highest uncooled detectivity.
|An SEM of our first-generation cavity-coupled detector.||The device’s measured lower limits for detectivity.|
Current infrared work in our group aims to understand radiation noise in micro- and nano-cavities. We have recently solved a 60 year old mathematical problem in photon statistics that allows us to incorporate appropriate levels of Poisson and Bose-Einstein noise in micro-cavities, and we are currently developing a radiation background theory for them.